The importance of ensuring access to justice in the development and use of AI systems
AI systems should not be used to replace or undermine existing legal systems or the right to access justice, including the right to a fair trial and due process. Rather, efforts should be made to ensure that AI is used in ways that enhance access to justice and protect human rights.
AI can be used to improve access to justice, but there are also concerns about bias and fairness in AI-powered decision-making. Notably, AI technologies show potential in their ability to democratize legal services, including apps such as automated document preparation and online dispute resolution.
For instance, the A2J author deploys decision trees, a simple form of AI, to create document preparation tools for complex filings in public benefits law and housing law. On the other hand, JustFix, a non-profit, offers online tools that help with various landlord-tenant issues. In addition, applications have been developed to aid individuals with criminal expungement to prepare for unemployment hearings or divorce.
Notwithstanding, there are more reasons to become worried about AI’s potential impacts on access to justice. For instance, much of the industry’s current technology and breakneck momentum does not serve the interests of marginalized or underserved populations. Despite AI’s potential, critics warn that the current trajectory may risk aggravating existing disparities. The access to justice issues revolves around deeper structural inequities rather than just access to technology.
It is worth noting that AI development is predominantly unregulated and influenced by market forces that mainly favor influential, wealthy actors. As a matter of fact, tech companies are not developing AI for free, and they are interested in developing a product that attracts those who are able to pay for it.
Recommendations for ensuring access to justice in the development of AI systems
- Each AI used in the legal field should have its developers and promoters identified in order for the results to be integrated into the adversarial process.
- Supervised and unsupervised learning and the related moderation processes should be publicized to illustrate their existence.
- The sources’ knowledge and architecture and the nature of data should also be used for training algorithms. This means that the data preparation processes should be clarified to test the quality of the data and the AI foundations’ reliability.
- A supra-national registry should be developed to facilitate the deposition of information about who uses AI and why. In exchange, applicants should be granted certificates or rights of use. After the establishment of this knowledge ecosystem, the probable deployment and further reliance on an AI approach to justice can be considered, ensuring that AI serves all people equally and without favor.